Saif Rehman Nasir commited on
Commit
0f78fcf
1 Parent(s): a8ae450

Add RAG related data

Browse files
Files changed (2) hide show
  1. app.py +32 -1
  2. requirements.txt +3 -1
app.py CHANGED
@@ -1,11 +1,25 @@
1
  import gradio as gr
2
  from huggingface_hub import InferenceClient
 
 
 
 
 
 
 
 
 
3
 
4
  """
5
  For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
6
  """
7
  client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
8
 
 
 
 
 
 
9
 
10
  def respond(
11
  message,
@@ -23,6 +37,23 @@ def respond(
23
  if val[1]:
24
  messages.append({"role": "assistant", "content": val[1]})
25
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
26
  messages.append({"role": "user", "content": message})
27
 
28
  response = ""
@@ -45,7 +76,7 @@ For information on how to customize the ChatInterface, peruse the gradio docs: h
45
  demo = gr.ChatInterface(
46
  respond,
47
  additional_inputs=[
48
- gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
49
  gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
50
  gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
51
  gr.Slider(
 
1
  import gradio as gr
2
  from huggingface_hub import InferenceClient
3
+ import os
4
+
5
+ from pinecone import Pinecone, ServerlessSpec
6
+ from sentence_transformers import SentenceTransformer
7
+
8
+
9
+
10
+
11
+
12
 
13
  """
14
  For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
15
  """
16
  client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
17
 
18
+ embedding_model = SentenceTransformer('sentence-transformers/all-MiniLM-L6-v2')
19
+
20
+ pinecone_client = Pinecone(api_key = os.getenv('PINECONE_API_KEY'))
21
+
22
+ index = pinecone_client.Index("movies")
23
 
24
  def respond(
25
  message,
 
37
  if val[1]:
38
  messages.append({"role": "assistant", "content": val[1]})
39
 
40
+ # encode user query
41
+ encoded_query = embedding_model.encode(message)
42
+
43
+ # retrieve most relevant movie from vector db
44
+ matches = index.query(
45
+ vector= encoded_query.tolist(),
46
+ top_k=1,
47
+ include_metadata = True
48
+ )
49
+
50
+ # movie which is most similar
51
+ retrieved_data = matches['matches'][0]['metadata']['title']
52
+
53
+ # Add as context to LLM
54
+ messages.append({"role":"user", "content": retrieved_data})
55
+
56
+
57
  messages.append({"role": "user", "content": message})
58
 
59
  response = ""
 
76
  demo = gr.ChatInterface(
77
  respond,
78
  additional_inputs=[
79
+ gr.Textbox(value="You are a movie recommender named Exodia. You are extremely reliable. You always mention your name in the beginning of conversation. You will provide me with answers from the given info.", label="System message"),
80
  gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
81
  gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
82
  gr.Slider(
requirements.txt CHANGED
@@ -1 +1,3 @@
1
- huggingface_hub==0.22.2
 
 
 
1
+ huggingface_hub==0.22.2
2
+ pinecone-client==5.0.0
3
+ sentence_transformers